This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Journal of Research in Personality 45 (2011) 285–296
Contents lists available at ScienceDirect
Journal of Research in Personality
journal homepage: www.elsevier.com/locate/jrp
Big Five personality dimensions in Italian and Dutch adolescents: A
cross-cultural comparison of mean-levels, sex differences, and associations
with internalizing symptoms
Theo A. Klimstra a,b,⇑,1, Elisabetta Crocetti c, William W. Hale III b, Alessandra Fermani c, Wim H.J. Meeus b
a
b
c
Catholic University Leuven, Belgium
Utrecht University, The Netherlands
University of Macerata, Italy
a r t i c l e
i n f o
Article history:
Available online 12 March 2011
Keywords:
Cross-cultural
Measurement invariance
Personality
Adolescence
Five-factor model
Big Five
Anxiety
Depression
a b s t r a c t
In the present cross-national comparison, self-reported Big Five personality data on large samples of
Dutch (N = 1521) and Italian (N = 1975) adolescents were employed. Results suggest that the personality
of Dutch and Italian adolescents can be described by the same Big Five traits, but that these might have
slightly different meanings to the Dutch and Italian adolescent respondents. Supplementary analyses
uncovered that sex differences are largest among Italian adolescents. Further comparisons reveal subtle
cross-national differences in personality–psychopathology relationships, with stronger associations of
Emotional Stability with depression for Italian when compared to Dutch adolescents. Results underscore
that cross-national comparisons of personality may be alluring to use in research, however the findings of
these comparisons should be interpreted with caution.
Ó 2011 Elsevier Inc. All rights reserved.
1. Introduction
In the previous two decades, there has been a growing consensus that the higher-order structure of personality can be described
with five broad dimensions (Caspi, Roberts, & Shiner, 2005). These
so-called ‘‘Big Five’’ dimensions are Extraversion (activity and
dominance in social situations), Agreeableness (investment in
maintaining positive and reciprocal relationships with others),
Conscientiousness (planful, organized, and responsible behavioral
tendencies), Emotional Stability (ability to deal with negative
emotions in an effective manner), and Openness (curiosity and
creativity) (Caspi et al., 2005; McCrae & John, 1992). The Big Five
dimensions have been claimed to be universally replicable in
adults (McCrae & Costa, 1997) as well as in adolescents (de Fruyt
et al., 2009), although some researchers have expressed doubts
with regard to this universality claim (e.g., de Raad et al., 2010;
Peabody & de Raad, 2002).
Still, the claim regarding the universality of the Big Five has
inspired several cross-cultural comparisons. Several studies,
⇑ Corresponding author at: Department School Psychology and Child and
Adolescent Development, Catholic University Leuven, Tiensestraat 102, bus 3717,
3000 Leuven, Belgium.
E-mail address: [email protected] (T.A. Klimstra).
1
The first author is Postdoctoral Researcher at the Fund for Scientific Research
Flanders (FWO).
0092-6566/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jrp.2011.03.002
including dozens of Western and developing countries (McCrae,
2001; Schmitt et al., 2007), have found substantial cross-cultural
differences in self-reported mean-levels of personality traits in college students and adults. For example, the study by McCrae (2001)
revealed that mean-level differences between countries were
related to Hofstede’s (2001) well-known dimensions of culture.
Power distance (i.e., acceptance of status differences) was
negatively related to Extraversion, and positively related to Conscientiousness, uncertainty avoidance (i.e., engagement in activities
to minimize threats in ambiguous situations) was negatively associated with Emotional Stability and Agreeableness, and there were
positive associations of individualism (i.e., assertiveness and need
for autonomy) with Extraversion and Openness. Only masculinity-feminity (i.e., the degree of sex-role differentiation which is
highest in masculine cultures and lowest in feminine cultures)
was unrelated to mean-level differences between countries. In a
follow-up study that included additional cultures, Allik and McCrae
(2004) reanalyzed McCrae’s (2001) data and related the findings
(2001) to geographical regions. Allik and McCrae (2004) demonstrated that individuals from European and American cultures tend
to report higher levels of Openness and Extraversion, and lower
levels of Agreeableness when compared to individuals from Asian
and African cultures. This line of research was extended by Schmitt
et al. (2007) whom distinguished somewhat more specific
geographical regions. Comparatively speaking, they found that
East-Asians tend to report the lowest levels on all Big Five
Author's personal copy
286
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
dimensions, that Africans reported the highest levels of
Conscientiousness, Agreeableness, and Emotional Stability, while
South-Americans reported the highest scores on Openness. Within
Europe, they found that Southern Europeans reported higher levels
of Agreeableness and Conscientiousness, but lower levels of
Emotional Stability when compared to Western Europeans.
However, it should be noted that some caution is warranted in
conducting analyses on a geographical region level, as McCrae
(2001) demonstrated that there can be substantive differences
between neighboring countries. For example, in McCrae’s study
(2001) the Dutch were among the highest ranking countries in
Emotional Stability, whereas their neighbors (Belgium and
Germany) were ranked much lower on this trait. As a result,
Schmitt et al.’s (2007) findings on Europeans only partly replicate
previous work by McCrae (2001), as McCrae also found that Western Europeans (e.g., the Dutch) reported higher levels of Emotional Stability when compared to Southern Europeans (e.g.,
Italians and Spaniards). For Agreeableness, McCrae’s findings contradict Schmitt et al.’s findings, as Western Europeans were more
agreeable than Southern Europeans in McCrae’s study, while the
inverse pattern (with Southern Europeans being more agreeable
than Western Europeans) was found in Schmitt et al.’s study. Thus,
findings concerning the geographical distribution of personality
trait-scores in Europe have not been all that consistent until now.
Sex differences in self-reported personality traits have also been
compared across cultures. Generally, sex differences appeared to
be larger in more economically prosperous (i.e., European and
North-American) countries when compared to less economically
prosperous (i.e., African and Asian) countries (Costa, Terracciano,
& McCrae, 2001; Schmitt, Realo, Voracek, & Allik, 2008). In these
two studies, the pattern of sex differences across cultures was
compared with Hofstede’s (2001) cultural dimensions. Only one
of these cultural dimensions (i.e., individualism–collectivism)
was associated with the magnitude of sex differences, indicating
that sex differences were larger in more individualistic countries.
Intuitively, one might expect sex differences to be larger in more
masculine cultures. However, Hofstede’s (2001) masculinity-femininity dimension was not associated with sex differences in the
studies of Costa et al. (2001) and Schmitt et al. (2008), suggesting
that countries with more differentiated sex roles are not necessarily characterized by larger sex differences in personality traits.
All the aforementioned cross-cultural comparisons focused only
on adult and college student samples, but recently it has been suggested that these findings can also be replicated among adolescent
samples (McCrae et al., 2010). However, instead of simply using
adolescent self-reported data, as has been done in adult and college
student samples, the study by McCrae et al. (2010) instead relied on
third-person ratings of the adolescents’ personalities. More specifically, these raters were assigned a target (i.e., an adolescent), whose
personality they had to rate. However, a similiarly recent work by
Wood, Harms, and Vazire (2010) has demonstrated that reports
about someone else’s personality might be as informative about
the personality of the perceiver (the third-person providing the
ratings) as they are about the personality of the target (the person
being rated). In other words, the value of McCrae et al. (2010)
replication of adult and college student findings with an adolescent
sample may be somewhat limited. Therefore, it would appear that
adolescent self-reported personality data is needed in addition to
third-person judgment based data, in order to obtain more insight
into cross-cultural differences in adolescent personality traits. However, as of the present, we are unaware of any previous cross-cultural
studies that have employed adolescent personality self-reports.
A lack of studies using adolescent self-reports is not the only
thing that is missing in regard to cross-cultural research. Extensive
literature regarding the linkages between Big Five personality
traits and psychopathology in adults (Kotov, Gamez, Schmidt, &
Watson, 2010) and accumulating knowledge regarding these associations in adolescents (Tackett, 2006) have revealed substantive
negative associations of psychopathology with Emotional Stability
and Extraversion. Additionally, negative associations have been
found with Agreeableness and Conscientiousness, albeit weaker,
but still significant. Again, despite the substantive number of studies on personality–psychopathology linkages, we are unaware of
cross-cultural comparisons of such associations. Because it is well
known that people from different countries express psychopathological symptoms in different ways (Lewis-Fernandez & Kleinman,
1994), and personality traits also appear to have a slightly different
meanings to persons across cultures (Nye, Roberts, Saucier, & Zhou,
2008), it seems obvious that personality–psychopathology linkages
found in one culture may not necessarily apply to other cultures.
While cross-cultural comparisons of personality–psychopathology
associations could considerably add to our understanding of the
meaning that specific personality traits and/or specific psychopathological symptoms may have across the globe, certain conditions to such cross-cultural comparisons must be applied.
Specifically, before cross-cultural comparisons of any kind can be
conducted, one should first perform several tests to rule out that
statistical artifacts are the most likely cause of one’s results.
1.1. Methodological issues with cross-cultural comparisons
In interpreting cross-cultural mean-level difference findings it
is important to be cautious for several reasons (Perugini & Richetin,
2007). One potential source of error that can influence the validity
of cross-cultural comparisons is the ‘‘frame-of-reference effect’’
(Heine, Lehman, Peng, & Greenholtz, 2002). The frame-of-reference
effect is that a person will compare him or herself to reference persons which usually happen to be the one’s in their direct proximity
(i.e., people from their own culture). Because of this effect, an
Italian might compare him or herself to other Italians when providing personality ratings, whereas a Dutch person might compare
him or herself to other Dutch persons. It is perhaps needless to
say that the frame-of-reference effect would obscure the validity
of mean-level differences between two cultures (in this case, between the Italians and the Dutch).
However, the most important reason to be cautious with crosscultural comparisons is that one needs to make sure that the
instrument that is used measures the construct in similar ways
across cultures. In other words, measurement invariance for the
instrument needs to be established before cross-cultural differences can be interpreted (e.g., Cheung & Rensvold, 2002). It is for
this reason that Exploratory and Confirmatory Factor Analytic
approaches are widely used (Caprara, Barbaranelli, Bermudez,
Maslach, & Ruch, 2000). The aforementioned cross-cultural studies
(McCrae, 2001; McCrae et al., 2010; Schmitt et al., 2007) have
mainly relied on Exploratory Factor Analysis (EFA) techniques to
establish measurement invariance. However, it has been argued
that EFA is mainly useful to test whether there is bias at the configural level (i.e., examining whether the number of factors and the
pattern of factor loadings is roughly equivalent in different groups
or cultures; Vandenberg & Lance, 2000), but is somewhat limited in
its potential to detect additional forms of bias. In order to overcome these limitations, Item Response Theory (IRT) methods and
Confirmatory Factor Analyses (CFAs) have been recommended to
test for two additional types of measurement invariance (i.e., metric and scalar invariance). In order to first establish metric invariance, one needs to check whether constraining factor loadings to
be equal across groups affects model fit. Secondly, scalar invariance
is tested by examining whether model fit is affected by constraining intercepts of latent factor indicators (e.g., items) to be equal
across groups/cultures. Together metric and scalar invariance tests
are useful in order to check for systematic response bias
Author's personal copy
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
(Vandenberg & Lance, 2000). That is, individuals from different
cultures might rate themselves differently on an item solely
because they interpret it differently, and not because of true
mean-level differences. For example, the item ‘‘neat’’ may be more
related to being orderly in the Netherlands, and perhaps more to
having your hair properly cut in Italy (although we are aware that
this particular example is somewhat stereotypical). To put it more
technically, one needs to demonstrate that individuals from different cultures with a similar mean-level on a latent factor, also have
a similar pattern of scores across the indicators of this latent factor
(i.e., the items).
Previous work with IRT-methods revealed that it is difficult to
demonstrate full metric and scalar invariance in commonly used
personality inventories, such as the Multidimensional Personality
Questionnaire (Johnson, Spinath, Krueger, Angleitner, & Riemann,
2008), the Trier Personality Inventory (Ellis, Becker, & Kimmel,
1993), and the NEO Personality Inventory (Huang, Church, &
Katigbak, 1997). However, IRT-methods are only sporadically used
because they are only recommended for very large samples of
more than 1000 cases (Stark, Chernyshenko, & Drasgow, 2006). A
good alternative for IRT-methods are Confirmatory Factor Analyses
(CFAs), as these analyses can be used to detect the same forms of
bias, but are less demanding in terms of the required sample size.
Unfortunately, CFAs have also rarely been used in cross-cultural
research on personality. One exception is a study by Nye et al.
(2008). Similar to the aforementioned IRT-studies, they were also
unable to establish full measurement equivalence. Thus, establishing full measurement equivalence seems hard to achieve, even
though it is a necessity for assuring that cross-cultural differences
reflect true mean-level differences, and cannot be solely attributed
to different interpretations of specific items.
1.2. The present study
To the best of our knowledge, the current study will be the first
to provide a cross-cultural comparison of adolescent self-reported
personality traits. To pursue this goal, large samples of Dutch
(N = 1521) and Italian (N = 1975) adolescents will be employed.
However, before this goal can be pursued, a prerequisite is the
establishment of measurement invariance, as aforementioned.
For that purpose, we will examine configural, scalar, and metric
invariance of a 30-item version of Goldberg’s Big Five questionnaire (Gerris et al., 1998; Goldberg, 1992) that the Dutch and
Italian adolescent samples completed. If measurement invariance
can be established, we will be able to pursue the following three
research goals: (1) comparing mean-levels of adolescent selfreported Big Five personality traits of Dutch and Italian adolescents, (2) compare sex differences in these traits across the two
cultures, and (3) examine cross-cultural differences in personality–psychopathology linkages. In all analyses, we will distinguish
among early and middle adolescents, because previous studies
found considerable age-related changes in Big Five traits among
adolescents (Klimstra, Hale, Raaijmakers, Branje, & Meeus, 2009).
Hence, the first goal of this study will be to assess mean-level
cross-cultural differences in adolescent self-reported Big Five
personality traits. It is reasonable to expect personality differences
between Dutch and Italian adolescents, as they have been shown
to differ on various variables that are related to personality. For
example, a report of the World Health Organization (WHO) found
that Italian adolescents report having more trouble communicating
with their parents when compared to Dutch adolescents (Currie
et al., 2008). Furthermore, Italian adolescents report having fewer
friends when compared to Dutch adolescents. In addition to these
cross-cultural differences found in the WHO study, Crocetti and colleagues found that Italian adolescents report having higher levels of
anxiety (Crocetti, Hale, Fermani, Raaijmakers, & Meeus, 2009) and a
287
less mature identity (Crocetti, Schwartz, Fermani, & Meeus, 2010)
when compared to Dutch adolescents. Adolescent personality traits
have been shown to be related to all the aforementioned. That is,
specific adolescents personality traits have been shown to be
associated with parent–child relationship quality (Agreeableness;
Branje, van Lieshout, & van Aken, 2004), the number of friends
one selects and the number of times one is selected as a friend
(Extraversion and Agreeableness, respectively; Selfhout et al.,
2010), anxiety (Emotional Stability; Krueger, 1999), and identity
(Agreeableness and Conscientiousness; Crocetti, Rubini, & Meeus,
2008; Luyckx, Soenens, & Goossens, 2006). As personality traits
have been shown to be related to so many variables on which
cross-cultural differences have been demonstrated, such differences are also likely to emerge in adolescent personality traits. If
cross-cultural differences in mean-levels of personality traits are
found, the aforementioned studies suggest that Dutch adolescents
are likely to reflect more favorable (i.e., higher) personality trait
scores than Italian adolescents, especially for Extraversion, Agreeableness, Conscientiousness, and Emotional Stability. For Openness,
we are unsure of what to expect, as this trait has only weak linkages
with the variables on which Dutch and Italian adolescents tend to
differ.
The second goal of the study is to compare the magnitude of sex
difference in the Dutch and the Italian sample. Using different
questionnaires, sex differences in Big Five traits have been found
both in Dutch (e.g., Klimstra et al., 2009) and Italian adolescents
(e.g., Caprara, Barbaranelli, Pastorelli, & Cervone, 2004). Although
it is less than ideal to compare results obtained with different
questionnaires, a glance at these studies seems to suggest larger
sex differences among Italian adolescents when compared to
Dutch adolescents. In the current study, we aim to provide a more
appropriate cross-cultural test by comparing sex differences in
adolescent self-reported personality traits as examined by a Dutch
and an Italian version of the same Big Five questionnaire.
The third and final goal of the current study is to compare personality–psychopathology linkages in Italian and Dutch adolescents. We will focus on two types of commonly experienced
psychopathological symptoms of adolescents: depressive symptoms and the Generalized Anxiety Disorder (GAD) symptom of
worry. These two forms of psychopathology have previously been
found to be strongly intertwined with personality traits (i.e., Emotional Stability, Extraversion, and Conscientiousness) of Dutch adolescents (e.g., Hale, Klimstra, & Meeus, 2010; Klimstra, Akse, Hale,
Raaijmakers, & Meeus, 2010), but we are unaware of studies examining associations of Big Five personality traits with the GAD symptom of worry and depressive symptoms in Italian samples. A study
among late childhood participants did however uncover linkages
between the Big Five traits (i.e., Emotional Stability, Extraversion,
Conscientiousness, and Openness) and a broad internalizing factor
comprised of depressive, anxiety, somatic, and obsessive symptoms (Barbaranelli, Caprara, Rabasca, & Pastorelli, 2003). Therefore,
it seems reasonable to expect that the GAD symptom of worry and
depressive symptoms will be associated with roughly the same Big
Five personality traits (i.e., Emotional Stability, Extraversion, and
Conscientiousness) in Dutch and Italian adolescents. However,
the present study will provide a more detailed perspective on these
linkages, as depressive symptoms and the GAD symptom of worry
will be considered separately.
2. Method
2.1. Participants
The Italian sample consisted of 1975 adolescents (902 boys and
1073 girls) attending various junior high and high schools in the
Author's personal copy
288
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
east-central region of Italy. Participants ranged in age from 11 to
19 years (Mage = 14.5, SD = 2.4). Two age groups were represented
in the sample: an early adolescent group (aged 11–14 years) of
1050 adolescents (Mage = 12.5 years, SD = 1.0) and a middle adolescent group (aged 15–19 years) of 925 adolescents (Mage =
16.8 years, SD = 1.2).
The Dutch sample consisted of 1521 adolescents (706 boys and
815 girls) attending various junior high and high schools in the
province of Utrecht in the Netherlands. Participants ranged in
age from 11 to 19 years (M = 14.2; SD = 2.2). Two age groups were
represented in the sample: an early adolescent group (aged 11–
14 years) of 880 adolescents (M age = 12.3 years, SD = 0.6) and a
middle adolescent group (aged 15–19 years) of 641 adolescents
(M age = 16.7 years, SD = 0.8).
To ensure that cross-national differences were not confounded
with ethnicity, we only focused on indigenous Italian and Dutch
adolescents. Additionally, adolescents in both countries attended
school full-time, and they were comparable in terms of years and
type of education. In fact, while in both countries there is no differentiation within the junior high school system, high schools are
differentiated in various tracks (from the highest level represented
by schools who prepare pupils for university attendance to the
lowest level represented by vocational schools). All these tracks
were represented in the Italian and Dutch samples.
2.2. Procedure
Prior to initiating the study, we obtained permission from the
school principals to administer questionnaires during class time.
Parents were provided with written information about the research and were asked for their consent for the adolescent to participate. After we received parental permission, students were
informed about the study and asked whether they wished to participate. Approximately 99% of the approached students chose to
participate. Interviewers then visited the schools and asked adolescents to fill out the questionnaire packet. This procedure was followed for both the Italian and Dutch adolescent samples.
2.3. Measures
2.3.1. Personality
A shortened version of Goldberg’s Big Five questionnaire (Gerris
et al., 1998; Goldberg, 1992) was used. Participants were asked to
rate 30 items (6 items for each factor) on a seven-point scale, ranging from 1 (does not apply to me at all) to 7 (applies to me very well).
Sample items include: talkative (Extraversion), sympathetic
(Agreeableness),
systematic
(Conscientiousness),
nervous
(Emotional Stability), and versatile (Openness to Experience).
Cronbach’s alphas were .70 and .82 for Extraversion, .71 and .85
for Agreeableness, .72 and .83 for Conscientiousness, .72 and .81
for Emotional Stability, and .65 and .75 for Openness to Experience,
in the Italian and Dutch samples, respectively.
2.3.2. Depression
The Children’s Depression Inventory (CDI; Kovacs, 1985, 1988;
Timbremont & Braet, 2002) was used to measure sub-clinical
depressive symptoms. The CDI consists of 27 items, each
responded to on a three-point scale: 1 (false), 2 (a bit true), and 3 (very
true). A sample item is ‘‘I am sad all the time’’. Cronbach’s alphas were
.88 and .92 in the Italian and Dutch samples, respectively.
2.3.3. Generalized anxiety symptoms
The Generalized Anxiety Symptoms (GAD) subscale from the
Screen for Child Anxiety Related Emotional Disorders (SCARED;
Birmaher et al., 1997; Crocetti et al., 2009; Hale, Raaijmakers,
Muris, & Meeus, 2005) was used to assess the generalized anxiety
disorder symptom of worry. The GAD scale consists of 7 items
scored on a three-point scale: 1 (almost never), 2 (sometimes),
and 3 (often). A sample item is: ‘‘I worry about whether others will
like me’’. Measurement invariance for this measure has been established for early and middle adolescent Dutch and Italian boys and
girls in a previous study (Crocetti et al., 2009). Cronbach’s alphas
were .76 and .86 in the Italian and Dutch samples, respectively.
3. Results
3.1. Preliminary analyses: testing measurement equivalence for
personality
We tested for three types of measurement equivalence (configural, metric, and scalar invariance) of personality by employing
Confirmatory Factor Analysis (CFA) in Mplus 4 (Muthén & Muthén,
2007) using Maximum Likelihood (ML) estimation. Tests for configural invariance are used to establish whether a model that yields
an adequate fit in one sample also yields an adequate fit in another
sample. No formal model comparisons can be run with regard to
configural invariance, as this would imply comparing two nonnested models.
To test for metric and scalar equivalence, we compared the fit of
multigroup CFA models (with Dutch early adolescent boys, Dutch
middle adolescent boys, Dutch early adolescent girls, Dutch middle
adolescent girls, Italian early adolescent boys, Italian middle adolescent boys, Italian early adolescent girls, and Italian middle adolescent girls as groups) without constraints (i.e., models with
metric or scalar variance) to constrained models (i.e., models with
metric or scalar invariance). For such model comparisons, the use
of multiple criteria has been advocated by Vandenberg and Lance
(2000), as different criteria can provide information on different
sources of potential model misspecification. Because the v2-statistic is well known to be overly sensitive to sample size and model
complexity (e.g., Cheung & Rensvold, 2002), we relied on two other
commonly used fit indices: the delta (D) Comparative Fit Index
(CFI) and the delta (D) Root Means Square Error of Approximation
(RMSEA). We concluded that there was measurement equivalence
if DCFI was smaller than .010 and DRMSEA was smaller than .015
(Chen, 2007). Absolute fit indices of the various models were also
considered, with CFIs of .90 and larger and RMSEAs of .08 and
smaller considered satisfactory (Kline, 2005).
For CFAs, using items as indictors of latent factors can lead to
overly complex models with a large number of parameters to be
estimated. In addition, it has been argued that the optimal number
of indicators for latent factors is three as it leads to a just-identified
model, whereas fewer indicators lead to an under-identified model
and more than three indicators yield an over-identified model (Little, Cunningham, Shahar, & Widaman, 2002). To reduce the number of indicators of a latent factor to the optimal number of three
and thereby reduce model complexity, it has been recommended
to use parcels consisting of multiple items instead of using individual items (e.g., Marsh & Hau, 1999). We used the well-established
item-to-construct balance parceling method (Little et al., 2002) to
create three two-item parcels for each Big Five trait resulting in a
total of 15 parcels. These parcels were used as input for CFAs in
which we tested for configural, metric, and scalar invariance. In
all models, the five latent factors representing the Big Five personality traits were allowed to correlate with each other.
To test for configural invariance, we first examined a simplestructure five-factor model in which the latent factors were
allowed to correlate with one another for the Dutch sample. This
model is depicted in Fig. 1. This model yielded a reasonable fit,
although the RMSEA was slightly higher than the .08 benchmark
advocated by Kline (2005) (see Table 1).
Author's personal copy
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
Fig. 1. Estimated Confirmatory Factor Analysis model for establishing configural
invariance.
Table 1
Model fit statistics of a five-factor model of personality in Dutch and Italian early and
middle adolescent boys and girls.
Dutch
Males
Females
Early Adolescents
Middle Adolescents
Italian
Males
Females
Early Adolescents
Middle Adolescents
v2
df
CFI
RMSEA
RMSEA (90% C.I.)
1083.55***
644.414***
567.665***
688.757***
481.546***
883.44***
383.915***
600.701***
457.560***
521.277***
77
77
77
77
77
77
77
77
77
77
.910
.897
.916
.904
.920
.904
.913
.889
.908
.894
.093
.102
.088
.095
.091
.073
.066
.080
.069
.079
.088–.098
.095–.110
.082–.095
.089–.102
.083–.098
.069–.078
.060–.073
.074–.086
.063–.075
.073–.085
Note. df = degrees of freedom; CFI = Comparative Fit Index; RMSEA = Root Mean
Square Error of Approximation.
⁄
p < .05.
⁄⁄
p < .01.
***
p < .001.
289
As Table 1 shows, the same model that we applied to the Dutch
sample also yielded a reasonable fit in the Italian sample.
Therefore, we established configural invariance for nationality.
We proceeded with tests for configural invariance in all gender
and age groups. These tests revealed that the model provided a
good fit for Dutch girls, Dutch early adolescents, Dutch middle adolescents, Italian boys, and Italian early adolescents. The models for
Dutch males, Italian females, and Italian middle adolescents
yielded a slightly less optimal fit (see Table 1). However, the fit statistics still seem to indicate that there is reasonable evidence for
configural invariance. Therefore, we could proceed with tests for
metric and scalar invariance.
We first tested for metric invariance by running two multigroup
simple-structure CFAs with eight groups (i.e., Dutch early adolescent boys, Dutch middle adolescent boys, Dutch early adolescent
girls, Dutch middle adolescent girls, Italian early adolescent boys,
Italian middle adolescent boys, Italian early adolescent girls, and
Italian middle adolescent girls) as groups. Because the fit of a model in which factor loadings were freely estimated for each of the
groups [metric variance model: v2 (617) = 2570.583 (p < .001),
CFI = .901, RMSEA = .085 (90% C.I. = .082–.089)] fitted the data as
well as a model in which these factor loadings were constrained
to be equal for all groups [metric invariance model: v2 (687) =
2828.920 (p < .001), CFI =.892, RMSEA = .084 (90% C.I. = .081 .088); DCFI = .009, DRMSEA = .001], we concluded that our Big
Five measure was metrically invariant for the eight groups involved. Therefore, we could proceed to the scalar invariance tests.
Our baseline model for scalar invariance tests was the eightgroup model in which factor loadings were constrained to be equal
for all groups, but in which intercepts were freely estimated. This
baseline model (i.e., the scalar variance model, which is a simplestructure model) was compared to a model in which intercepts
were constrained to be equal for all groups (i.e., the scalar
invariance model, which is a mean-structure model). The scalar
invariance model [v2 (757) = 4610.708 (p < .001), CFI = .805,
RMSEA = .108 (90% C.I. = .105–.111)] had a fit that was much worse
than the scalar variance model (DCFI = .087, DRMSEA = .024). Consequently, we concluded that there was no overall scalar invariance. However, we did proceed to test whether we could
establish scalar invariance for specific pairs of gender and age
groups within and across cultures. The results of these comparisons appear in Table 2.
Table 2 indicates that scalar invariance was established within
the Dutch sample in almost all instances. Comparisons between
Dutch middle adolescent boys and girl should, however, be interpreted with some caution as DCFI (.011) was just above the .010
benchmark. Within the Italian sample, the situation was the complete opposite, as the only comparison that signified scalar invariance was the comparison between Italian middle adolescent boys
and girls. In all other instances, group comparisons within the Italian culture should be interpreted cautiously.
There was no evidence for cross-cultural scalar invariance.
Thus, cross-cultural mean-level differences of Big Five traits may
be interpreted (because there was configural and metric invariance), but only cautiously (because of a lack of scalar invariance).
As there was metric invariance for sex groups but no scalar invariance within the Italian sample, comparisons of sex differences
across cultures and mean comparisons within the Italian sample
should also be interpreted with some caution.
Cross-cultural comparisons of associations of personality traits
with other variables (i.e., problem behavior symptoms) ideally require metric invariance, which we managed to establish across all
groups. Therefore, these comparisons can be readily interpreted. In
fact, between-group differences in associations of personality with
problem behavior symptoms could even provide some insights
into the (slight) differences in meaning a specific personality trait
Author's personal copy
290
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
Table 2
Scalar invariance tests for nationality, gender, and age.
Scalar variance
Scalar invariance
2
v
***
Overall gender by nationality
Dutch boys vs Dutch girls
Italian boys vs Italian girls
Dutch boys vs Italian boys
Dutch girls vs Italian girls
Overall age by nationality
Dutch early vs Dutch middle
Italian early vs Italian middle
Dutch early vs Italian early
Dutch middle vs Italian middle
Dutch early boys vs Dutch early girls
Dutch mid boys vs Dutch mid girls
Dutch early boys vs Dutch mid boys
Dutch early girls vs Dutch mid girls
Italian early boys vs Italian early girls
Italian mid boys vs Italian mid girls
Italian early boys vs Italian mid boys
Italian early girls vs Italian mid girls
2354.093
1254.765***
1015.336***
1106.512***
1207.157***
2331.162***
1261.781***
989.550***
1257.135***
1015.531***
879.772***
569.000***
789.853***
692.858***
576.982***
640.844***
484.591***
711.619***
df
CFI
RMSEA (90% C.I.)
v2
df
CFI
RMSEA (90% C.I)
338
164
164
164
164
338
164
164
164
164
164
164
164
164
164
164
164
164
.896
.904
.897
.895
.901
.899
.904
.901
.896
.908
.890
.920
.888
.910
.899
.884
.910
.883
.083
.094
.073
.085
.082
.082
.094
.071
.083
.081
.100
.088
.105
.089
.069
.079
.066
.079
3894.098***
1337.788***
1116.132***
1706.516***
1946.843***
3886.417***
1365.145***
1134.960***
2173.703***
1512.983***
913.111***
634.868***
844.480***
764.585***
660.327***
671.581***
532.629***
831.314***
368
174
174
174
174
368
174
174
174
174
174
174
174
174
174
174
174
174
.819
.897
.885
.830
.831
.822
.896
.884
.809
.855
.886
.909
.881
.900
.881
.879
.899
.859
.105
.094
.074
.105
.104
.105
.095
.075
.109
.099
.098
.091
.104
.091
.073
.079
.068
.084
(.080, .086)
(.089, .098)
(.068, .077)
(.080, .089)
(.078, .086)
(.079, .085)
(.089, .099)
(.067, .076)
(.079, .087)
(.077, .086)
(.093, .016)
(.080, .093)
(.098, .112)
(.082, .096)
(.063, .075)
(.073, .086)
(.059, .073)
(.073, .085)
(.102–.108)
(.089–.099)
(.070–.078)
(.100–.109)
(.100–.108)
(.102–.108)
(.090–.100)
(.071–.079)
(.105–.113)
(.095–.104)
(.092–.105)
(.083–.099)
(.097–.112)
(.085–.098)
(.067–.079)
(.072–.085)
(.061–.074)
(.078–.090)
Note. Comparisons in which scalar invariance was established are printed in bold. df = degrees of freedom; CFI = Comparative Fit Index; RMSEA = Root Mean Square Error of
Approximation.
⁄
p < .05.
⁄⁄
p < .01.
***
p < .001.
conscientious and open to experience, but less emotionally stable
when compared to Dutch middle adolescent boys.
Sex differences in the Italian sample should be interpreted
cautiously, as we only managed to establish scalar invariance for
Italian middle adolescent boys and girls. That being said, it appears
from Table 3 that there were no sex differences in Extraversion
among Italian adolescents. Italian girls were more agreeable and
conscientious than Italian boys, both in early and middle adolescence. For Emotional Stability, Italian middle adolescent girls
reflected lower levels than Italian middle adolescent boys. For
Openness, there were no gender or age differences among Italians.
might have in these different groups. In other words, betweengroup differences in personality-problem behavior symptom linkages could potentially clarify why scalar invariance could not be
established.
3.2. Sex differences within cultures
Sex differences within cultures were inspected with a Multivariate Analysis Of Variance (MANOVA). In these analyses we accounted for age differences, as we examined sex differences
within age cohorts (i.e., Dutch early adolescent boys versus Dutch
early adolescent girl, Dutch middle adolescent boys versus Dutch
middle adolescent girls, Italian early adolescent boys versus Italian
early adolescent girls, and Italian middle adolescent boys versus
Italian middle adolescent girls). The results of these analyses appear in Table 3.
Within the Dutch sample, no sex differences were found with
regard to Extraversion (see Table 3). However, early adolescent
boys tended to be less agreeable than early adolescent girls. This
sex difference was replicated in middle adolescence, where Dutch
boys were again less agreeable when compared to Dutch girls.
Dutch middle adolescent girls also were significantly more
3.3. Mean-level differences across cultures
We assessed mean-level differences across cultures by comparing similar sex and age groups across cultures (e.g., Dutch early
adolescent boys versus Italian early adolescent boys, Dutch early
adolescent girls versus Italian early adolescent girls, Dutch middle
adolescent boys versus Italian middle adolescent boys, and Dutch
middle adolescent girls versus Italian middle adolescent girls) with
a MANOVA. It should be noted that these comparisons should be
interpreted very cautiously, because we did not establish
Table 3
Mean-levels of personality, depression and worry in early and middle adolescent Dutch and Italian boys and girls.
Early adolescents
Middle adolescents
Dutch
Italian
Boys (N = 434)
Personality
Ex
4.86
Ag
5.03
Co
4.10
ES
4.67
Op
4.46
(.87) cd
(.98) ab
(1.04) bc
(1.02) e
(.99) a
Int. problems
Dep
1.15 (.28)
GAD
1.33 (.35)
a
a
Girls (N = 446)
4.97
5.23
4.22
4.56
4.45
(.99) d
(.80) c
(1.01) cd
(.97) de
(.90) a
1.17 (.20)
1.35 (.30)
ab
a
Dutch
Boys (N = 520)
4.77
4.97
3.94
4.43
4.46
(1.20) cd
(.90) a
(1.13) ab
(1.05) cd
(1.02) a
1.36 (.28)
1.70 (.43)
c
c
Girls (N = 530)
4.65
5.41
4.27
4.22
4.63
(1.30) bc
(.84) d
(1.06) cd
(1.14) bc
(1.03) ab
1.35 (.27)
1.84 (.45)
c
d
Italian
Boys (N = 272)
4.65
5.16
4.08
4.51
4.60
(1.07) bc
(.86) bc
(1.06) abc
(.98) de
(.95) ab
1.19 (.26)
1.31 (.33)
ab
a
Girls (N = 369)
4.74
5.44
4.37
4.22
4.74
(1.14) cd
(.69) d
(1.13) d
(.96) bc
(.84) b
1.21 (.23)
1.48 (.43)
b
b
Boys (N = 382)
4.49
5.08
3.85
4.04
4.47
(1.16) ab
(.92) abc
(1.15) a
(1.12) b
(1.03) a
1.39 (.30)
1.77 (.43)
cd
cd
Girls (N = 543)
4.37
5.44
4.14
3.40
4.52
(1.40) a
(.78) d
(1.15) bcd
(1.06) a
(.94) a
1.43 (.31)
2.01 (.45)
d
e
Note. Ex = Extraversion, Ag = Agreeableness, Co = Conscientiousness, ES = Emotional Stability, Op = Openness, Int. problems = Internalizing Problems, Dep = Depression;
GAD = the Generalized Anxiety Disorder Symptom of Worry. Means of groups that share the same superscript are equal, means of groups with different superscripts are
significantly different (p < .05).
Author's personal copy
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
cross-cultural scalar measurement equivalence (see Table 2).
Descriptive statistics for these mean-level comparisons are
displayed in Table 3.
Table 3 suggests that Dutch girls (both early and middle adolescents) were more extraverted than their Italian counterparts,
whereas there were no cross-cultural mean differences for boys.
For Agreeableness, Dutch early adolescent girls exhibited higher
mean-levels than Italian early adolescent girls. Italian and Dutch
adolescents appear to be equally conscientious. With regard to
Emotional Stability, this was by no means the case as all Dutch
adolescent groups displayed higher levels on this trait than their
Italian counterparts. Finally, for Openness there was one cross-cultural difference as Dutch middle adolescent girls were more open
to experience than Italian middle adolescent girls.
3.4. Comparing age and sex differences in personality within cultures
across cultures
As a final mean-level analysis, we compared the magnitude of
sex differences across cultures with a MANOVA. Again, these results should be interpreted cautiously because of the lack of scalar
invariance across cultures. Descriptive statistics concerning these
analyses appear in Table 3.
First, a significant multivariate interaction effect of nationality
by sex (F(5, 3484) = 4.30; p = .001; partial g2 = .006) indicated that
the overall magnitude of gender differences differs between cultures. Univariate tests indicated that this interaction only applied
to Extraversion (F(1, 3488)= 7.23; p = .007; partial g2 = .002),
Agreeableness (F(1, 3488) = 7.46; p = .006; partial g2 = .002), and
Emotional Stability (F(1, 3488) = 9.64; p = .002; partial g2 = .003).
An inspection of Table 3 reveals that gender differences in these
traits were larger among Italian adolescents when compared to
Dutch adolescents.
To examine whether the sex by nationality effects were moderated by age group, we examined a three-way interaction effect of
nationality by sex by age group. The multivariate effect was not
significant (F(5, 3484) = 1.96; p = .081; partial g2 = .003), indicating
that sex by nationality interactions were comparable for early and
middle adolescents.
3.5. Comparing associations of Big Five personality traits with
internalizing problems across cultures, sex, and age groups
We ran a multigroup structural equation model in Mplus 4
(Muthén & Muthén, 2007) to examine the associations between
the Big Five personality traits and two distinct internalizing
problems (i.e., depressive symptoms and the Generalized Anxiety
Disorder symptom of worry) in the aforementioned eight nationality, gender, and age groups. Maximum Likelihood Robust (MLR)
estimation was used as MLR has been shown to be the most accurate estimator when the distribution of scores slightly deviates
from a normal distribution (Satorra & Bentler, 1994), which
turned out to be the case for the scores on our measures for
depression and worry. In the structural equation model, the Big
Five personality traits were the independent variables, whereas
depression and worry were the dependent variables. Associations
among independent variables, among dependent variables, and
predictive paths from the independent to the dependent variables
were all estimated. We used latent variables which were all indicated by three parcels. In our previous analyses, we had demonstrated that there was metric invariance for personality, whereas
a previous study on portions of the same datasets (Crocetti et al.,
2009) established metric invariance for our measure of the GAD
symptom of worry. Our depression measure was also found to
be metrically invariant, as fit differences between a metric
291
variance model and a metric invariance model were minimal
(DCFI was .004 and DRMSEA was .001). Therefore, we constrained
factor loadings of the parcels of each of the latent variables to be
equal across all eight groups. The resulting model is graphically
represented in Fig. 2.
The estimated model had an acceptable fit (v2 (1419) =
4345.989 (p < .001), CFI = .915, RMSEA = . 069 (90% C.I. = .066–
.071). To assess whether specific associations were similar or
different across culture, sex, and age groups, we compared the
confidence intervals of these associations for the different groups.
If the confidence intervals did not overlap, we concluded that there
was a significant between-group difference. All associations of Big
Five personality traits with depressive symptoms and the Generalized Anxiety Disorder symptom of worry, and the between-group
comparisons of these associations are depicted in Table 4.
3.5.1. Associations of Big Five personality traits with depressive
symptoms
Table 4 indicates that associations between depressive symptoms and Big Five personality in the eight distinguished groups
were predominantly negative. These associations will be discussed
trait-by-trait.
Extraversion was a significant negative predictor of depression
in middle adolescent Italian boys, and early and middle adolescent
Italian girls. However, confidence intervals indicated that these
associations did not significantly differ from the non-significant
associations in the other groups. Agreeableness was a negative predictor for depression in Dutch early adolescent boys and Italian
middle adolescent girls. Again, an inspection of the confidence
intervals indicated that these associations did not differ significantly from the non-significant associations found in the other
groups. Higher levels of Conscientiousness predicted lower levels
of depression in Dutch middle adolescent girls, and Italian middle
adolescent boys and girls. These associations did again not differ
from the non-significant associations found in the other groups.
Emotional Stability negatively predicted depression in six of the
eight distinguished groups. These associations only failed to reach
significance in Dutch early and middle adolescent boys. Associations of Emotional Stability were stronger in Italian early and middle adolescent boys and girls, when compared to Dutch early
adolescent girls and Dutch middle adolescent boys. Openness
was a significant positive predictor of depression in Dutch girls,
but in none of the other groups. In addition, this association was
no stronger than the non-significant associations between Openness and depression we found in the other groups.
3.5.2. Associations of Big Five personality traits with the GAD symptom
of worry
The GAD symptom of worry was associated with only three of
the Big Five personality traits, as Extraversion and Openness were
not associated with worry. In addition, there were substantive between-group differences in the associations we found.
Agreeableness was a significant negative predictor of worry,
but only in Dutch early adolescent boys. However, this negative
association did not significantly differ from the non-significant
associations found in all the other groups. Worry was positively
predicted by Conscientiousness in early adolescent Dutch girls,
but negatively predicted by this very same trait in middle adolescent Dutch girls. The difference between these two associations was significant. Finally, Emotional Stability was a
significant negative predictor of worry in all groups, except for
Dutch early adolescent boys. In addition, the negative association
of Emotional Stability with worry was significantly stronger in
Italian middle adolescent girls, when compared to Dutch early
adolescent girls.
Author's personal copy
292
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
Fig. 2. Estimated structural equation model with the Big personality traits of Extraversion (Ex), agreeableness (Ag), conscientiousness (Co), emotional stability (ES), and
openness (Op) as independent latent variables, and depressive symptoms (Dep) and the Generalized Anxiety Disorder symptom of worry (GAD) as dependent latent variables.
Circles represent latent variables, squares represent observed item parcels.
Table 4
Associations of Big Five personality traits with depressive symptoms and worry in Italian and Dutch adolescent boys and girls.
Early adolescents
Middle adolescents
Dutch
Italian
Boys
Ex ? depression
Ag ? depression
Co ? depression
ES ? depression
Op ? depression
Ex ? worry
Ag ? Worry
Co ? worry
ES ? worry
Op ? worry
.432
.282
.167
.166
.323
.064
.369
.072
.431
.251
Girls
ab
ab
ab
ab
.206
.109
.040
.243** a
.033
.071
.024 a
.221* a
.500*** a
.170
Dutch
Boys
.049
.064
.107
.483***
.131
.035
.016 ab
.021 ab
.482***
.085
Girls
b
ab
Boys
.126*
.126
.084
.503***
.040
.037
.044 ab
.075 ab
.591***
.044
b
ab
.127
.263**
.114
.077 a
.081
.015
.269** ab
.140 ab
.420** ab
.001
Italian
Girls
.068
.116
.165**
.314***
.147*
.040
.136 ab
.130* b
.527***
.110
Boys
ab
ab
.167*
.150
.160*
.464***
.033
.112
.027 b
.048 ab
.567***
.017
Girls
b
ab
.146**
.129*
.190***
.430***
.098
.002
.020 ab
.063 ab
.616***
.023
b
b
Note. Ex = Extraversion, Ag = Agreeableness, Co = Conscientiousness, ES = Emotional Stability, Op = Openness. Different superscripts within a line indicate significant (p < .05)
between-group differences in associations of personality traits with depressive symptoms and worry. Lines without superscript contain no significant between-group
differences.
*
p < .05.
**
p < .01.
***
p < .001.
4. Discussion
The purpose of the current study was to provide a first crossnational comparison on adolescent self-reported personality traits.
Italian adolescents were compared to Dutch adolescents in terms
of mean-levels, sex differences, and personality–psychopathology
linkages. In all analyses we also distinguished among early and
middle adolescents. However, a prerequisite for an appropriate
cross-national comparison is establishing that the instrument
one uses measures the construct of interest (i.e., personality) in
Author's personal copy
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
the same way in different cultures. In other words, before crossnational comparisons can be interpreted, measurement
equivalence must be established.
In the present study, we were able to establish configural
invariance. That is, we were able to show that personality as measured with a 30-item Dutch version of Goldberg’s (1992) Big Five
measure is appropriately described by roughly the same five-factor
structure for Italian and Dutch adolescents. Cross-national configural invariance of the Big Five personality traits had previously been
established for self-ratings of adult and college student samples
(McCrae, 2001; Schmitt et al., 2007), observer ratings on adult
and college student samples (McCrae, Terraciano, & 78 members
of the personality profiles of cultures project, 2005), and for observer ratings on adolescents (McCrae et al., 2010), but the present
study is the first to establish configural invariance for adolescent
Big Five self-reports.
However, it has recently been argued that only establishing
configural invariance is not sufficient for cross-national comparisons. In fact, it has been recommended to proceed with tests of
metric and scalar invariance (Nye et al., 2008). The former can be
considered a requirement for comparing associations among variables across groups, whereas the latter is advisable when one
wants to compare means (cf. Vandenberg & Lance, 2000). Both
metric and scalar invariance can be examined with Confirmatory
Factor Analyses. Such analyses can only be applied if the N for each
of the investigated countries is at least 100 to 200, but with more
complex models with large number of free parameters (such as a
five-latent-factor models with usually about 6 to 8 indicators for
each of the factors) larger samples (N > 200) may be necessary in
order to have models converge (Kline, 2005). It is perhaps because
such sample sizes are usually not obtained for each of the participating countries that most previous studies only examined configural invariance with exploratory factor analyses, and did not
test for metric and scalar invariance with Confirmatory Factor
Analyses.
In the present study, we did have a sufficiently large sample size
to proceed with tests for metric and scalar invariance. These tests
revealed that factor loadings were equivalent across all nationality,
sex and age groups, which indicates that we managed to established metric equivalence. As a result, associations between variables can be compared across these groups. We were, however,
unable to establish scalar invariance across all groups. Post-hoc
scalar invariance tests within and across cultures revealed that scalar invariance was only attainable within the Dutch sample. Therefore, mean comparisons between Dutch age and sex groups can be
readily interpreted, which underscores the validity of results obtained in previous studies that were conducted on the portion of
the Dutch participants that was followed longitudinally (e.g., Klimstra et al., 2009; Meenus, van de Schoot, Klimstra, & Branje in
press). Moreover, these findings reveal that expecting to establish
scalar invariance might not be as unrealistic as has been argued
in previous studies (e.g., Byrne, Shavelson, & Muthen, 1989).
Within the Italian sample, scalar invariance could not be established as imposing such constraints led to a significantly worse fit.
However, although the changes in the CFI passed the recommended threshold (>.010; Chen, 2007), the RMSEAs were similar.
Moreover, the CFI threshold was not violated in a dramatic way
(DCFIs for gender and age invariance tests within the Italian sample were .012 and .017). Thus, caution is warranted in interpreting
mean-level differences between sex and age groups within the Italian sample, but our findings would suggest that such differences
still can be interpreted, albeit carefully. The same cannot be said
of scalar invariance across cultures, as imposing these constraints
lead to a dramatic drop in the CFI and a substantive increase in
the RMSEA. Thus, when interpreting overall mean-level differences
or when comparing the magnitude of mean-level and sex
293
differences between the Dutch and Italian adolescents in the current study, one should bear in mind that these differences may
be due to the slightly different meanings these traits have for
Dutch and Italian adolescents. That is, the items included in the
scales of our personality measure are endorsed differently by Italian adolescents when compared to Dutch adolescents. Put more
simply, a high level of Agreeableness among Italian adolescents
may be mainly due to high scores on items like ‘‘cooperative’’,
whereas a high level of the same trait may be mainly due to high
scores on a slightly different item like ‘‘pleasant’’ in Dutch adolescents. We are not the first that have been unable to establish scalar
invariance in a cross-national comparison; several previous studies
(Ellis et al., 1993; Huang et al., 1997; Johnson et al., 2008; Nye
et al., 2008) that also applied these rigorous tests also were unable
to establish scalar invariance in several instances. Given that establishing scalar invariance was mainly problematic across cultures,
and less of an issue within cultures, the present study suggests that
comparing mean-levels across countries might be problematic.
Therefore, the findings of previous cross-national comparisons of
mean-levels in which scalar invariance has not been examined
(e.g., Allik & McCrae, 2004; McCrae, 2001; McCrae, Terraciano, &
78 members of the personality profiles of cultures project, 2005;
McCrae et al., 2010; Schmitt et al., 2007) should be interpreted cautiously. The same applies to the cross-cultural mean-level differences assessed in the present study, which will now be discussed.
Some evidence was found for cross-cultural mean-level differences on Agreeableness and Openness. However, these differences
only applied to early adolescent girls for Agreeableness and middle
adolescent girls for Openness. For Extraversion, cross-cultural differences were found for both early and middle adolescent girls,
with Dutch girls reporting higher levels of Extraversion than Italian
girls. Dutch adolescents tend to report having more friends than
Italians (Currie et al., 2008), and Extraversion is known to be associated with the number of friends one nominates (Selfhout et al.,
2010). Therefore, our findings do not come as a surprise. It is, however, unclear why these anticipated cross-cultural differences in
Extraversion only emerged for adolescent girls. One obvious reason
could be the lack of scalar invariance across cultures, and within
the Italian culture. This could contribute to different adolescent
groups endorsing the same items differently. As a result, Extraversion might have a slightly different meaning in boys when compared to girls, and in Italians when compared to Dutch. Thus,
before theorizing on these differences one first needs to rule out
that findings are due to a lack of scalar invariance. The largest
and most consistent cross-cultural differences emerged for
Emotional Stability, where all Dutch age and sex groups displayed
higher levels than their Italian counterparts. Although an overly
ambitious interpretation of these findings would not be in place
due to the lack of scalar invariance, our findings are in line with
our hypotheses based on cross-cultural mean-level differences in
anxiety (Crocetti et al., 2009) and highly consistent across sex
and age groups. Therefore, we feel we can cautiously state that
Italian adolescents appear to be less successful in dealing with negative emotions effectively, when compared to Dutch adolescents.
We also proceeded with an examination of sex differences within Italian and Dutch samples, and then compared the magnitude of
the sex differences across samples. In all these analyses, we distinguished among early and middle adolescents. Given the aforementioned problems with lacking scalar invariance, especially the
cross-national comparisons of the magnitude of sex differences
need to be interpreted cautiously. In Dutch adolescents, sex differences in Agreeableness mainly applied to early adolescents where
girls displayed higher scores than boys. All other prominent sex
differences among Dutch adolescents applied to middle adolescents, as middle adolescent girls were more conscientious and
open to experience, but less emotionally stable when compared
Author's personal copy
294
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
to middle adolescent boys. Sex differences in Agreeableness, Conscientiousness, and Emotional Stability were replicated among
Italian adolescents, but for Openness no sex differences were
found. In addition, sex differences in Agreeableness and Conscientiousness applied to both early and middle adolescents. Our findings for Italian adolescents with regard to Emotional Stability
perfectly replicate our findings obtained among Dutch adolescents,
as these finding also only applied to middle adolescents in our Italian sample.
A cross-national comparison of the magnitude of sex differences
revealed that, given cross-national differences in the endorsement
of specific items (signified by a lack of scalar invariance), sex differences were somewhat larger in Italy than in The Netherlands for
Extraversion, Agreeableness, and Emotional Stability. A first explanation for this finding would be that the items are endorsed in such
a way that trait-scores are closely linked to sex stereotypes in Italy,
whereas these traits may have a somewhat more neutral connation
in The Netherlands. Alternatively, sex differences may be larger in
Italy due to ‘‘machismo’’. Indeed, Hofstede’s (2001) study on
dimensions of culture revealed that Italy was one of the most masculine countries (rank 4 out of 53), whereas The Netherlands was
among the least masculine countries (rank 51 out of 53). On the
other hand, effect sizes of sex by nationality interactions were only
small and should therefore not be over-interpreted. In addition,
definite explanations of cross-national differences in the magnitude of sex differences can only be drawn once it has been established how the connotation of Big Five personality traits exactly
differs for Dutch and Italian adolescents. A cross-national comparison of personality–psychopathology linkages may provide some
preliminary insights into this issue.
Personality–psychopathology linkages were found to be quite
similar for Italian and Dutch adolescents with regard to depressive
symptoms, as all but one of the Big Five traits (i.e., Openness) were
negatively associated with depressive symptomatology in both
samples. Openness positively predicted depressive symptoms,
but only in Dutch middle adolescents. However, this marginally
significant predictive path was not significantly different from
the non-significant predictive paths found in the other sex, age,
and nationality groups. Cross-national differences did emerge for
linkages between Emotional Stability and depressive symptoms,
as these were stronger among Italian early adolescent girls than
among Dutch early adolescent girls, and stronger among Italian
middle adolescent boys when compared to Dutch middle adolescent boys. Thus, Emotional Stability appears to be a better predictor of depressive symptomatology among Italians than among the
Dutch. This may suggest that depressive symptoms are more likely
to be caused by stable dispositions such as personality among Italians, whereas external causes (e.g., life events, family climate, quality of peer relations) may be a more common cause of depressive
symptoms among Dutch. Future studies with a longitudinal focus
are needed to replicate our findings before further inferences can
be drawn. However, there would be profound implications for
the treatment of depression if the extent to which depressive
symptoms are caused by stable dispositions instead of more
changeable environmental factors indeed differs from one culture
to another.
For the Generalized Anxiety Disorder symptom of worry we
found fewer linkages with personality traits than for depressive
symptoms. Moreover, there were few cross-national differences
in these associations. Extraversion and Openness were unassociated with worry, whereas there was only a significant (negative)
association between Agreeableness and worry in Dutch middle
adolescent boys. Conscientiousness was significantly, but only
marginally, associated with worry in Dutch girls. However,
whereas Conscientiousness appeared to amount to worry in early
adolescence, it seemed to protect against worry in middle
adolescence. Thus, even though there was scalar invariance for
Conscientiousness for early adolescent girls when compared to
middle adolescent girls, the meaning of this trait still seems to differ slightly for these two age groups. While we do not want to
over-interpret the marginally significant positive versus negative
associations in Dutch early and middle adolescent girls, respectively, our findings do seem to suggest that establishing scalar
invariance does not yet mean that the meaning of a certain trait
is perfectly identical for two groups. Thus, we would recommend
considering associations with external variables such as problem
behavior in addition to measurement invariance tests when examining whether the connotation of traits is similar or different across
age, sex, and nationality groups.
Of all personality traits, Emotional Stability had the strongest
relations with worry. These associations appeared to be largely
equal across cultures. Despite these strong relations, a previous
study demonstrated that Emotional Stability and worry are two
distinct constructs (Hale et al., 2010). In that study of Dutch adolescents, more emotionally stable individuals were further found to
worry less on the subsequent measurement occasion, but less worried individuals were also found to be more emotionally stable on
the subsequent measurement occasion. Thus, worry and Emotional
Stability affected one another in a reciprocal fashion. An interesting
endeavor for future studies would be to investigate whether the
same developmental (i.e., longitudinal) processes with regard to
Emotional stability and worry can be replicated in different
cultures.
Overall, we found fewer linkages of personality traits with problem behavior symptoms than previous studies (Barbaranelli et al.,
2003; Hale et al., 2010; Klimstra et al., 2010), as there were very
few associations of Conscientiousness and Extraversion with worry
and depressive symptoms in the current study. In fact, only Emotional Stability was consistently associated with problem behavior
symptoms. This contrast with previous studies is probably caused
by the fact that we accounted for the associations among Big Five
personality traits. That is, the associations we present here are
correlations between problem behavior symptoms and the unique
aspect of each Big Five trait, corrected for the overlap between Big
Five traits. As such, our findings could suggest that previous
studies merely found associations of Conscientiousness and
Extraversion with problem behavior symptoms, because these
two traits partly overlap with Emotional Stability.
Another important observation regarding the personality–
psychopathology linkages in the current study is that the pattern
of associations of personality with depression seems largely similar
to the pattern of associations of personality with worry. There are,
however, subtle differences. The most prominent one refers to
Extraversion that seems to protect (Italian) adolescents against
depressive symptoms, whereas this trait does not serve such a role
with regard to worry. Subtle differences also emerge for Emotional
Stability. This trait appears to be stronger related to worry than to
depressive symptoms. Moreover, whereas there are relatively
consistent cross-national differences in associations of depressive
symptoms with Emotional Stability, such cross-national
differences do not seem to emerge with regard to the associations
of worry with Emotional Stability. Together, these subtle differences in linkages between personality with depressive symptoms
when compared to linkages between personality with worry suggest that a broad internalizing problem behavior factor might lead
to somewhat too general conclusions regarding personality–
psychopathology linkages. As such, our findings seem to favor
the specificity of research on psychopathology guided by DSMIV-TR criteria (Diagnostic and Statistical Manual of the Mental
Disorders, Fourth Edition, Text Revision; American Psychiatric
Association, 2000) in which the symptoms of disorders such as
Depression and the Generalized Anxiety Disorder symptom of
Author's personal copy
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
worry are examined separately, over approaches in which these
disorders are (together with other related disorders) collapsed into
one broad internalizing factor (e.g., Watson & Clark, 1984).
Despite some subtle cross-national differences, personality–
psychopathology linkages do not sufficiently clarify why crossnational scalar invariance could not be established in the current
study. Hence, it is unclear how the connotation of personality traits
differs in The Netherlands and Italy. Therefore, we would
recommend further research into cross-national differences in
the meaning of traits. In this regard, research on the motivational
underpinnings of personality traits might be very relevant.
Denissen and Penke (2008) have recently subsumed descriptions
of these motivational underpinnings, and provided their own
perspective on this topic. A brief glance at the overview and their
own ideas reveals that there is quite a bit of variation in descriptions of motivational underpinnings of the Big Five. For example,
Agreeableness has been described as a disposition to react cooperatively in resource conflict (Denissen & Penke, 2008), but also as a
trait facilitating positive family relationships (MacDonald, 1995,
1998). Similarly, Conscientiousness has been described in motivational terms as the tenacity of goal pursuit (Denissen & Penke,
2008), but also as trustworthiness and dependability (Hogan,
1996). It is hard to assess which description of motivational underpinning is the most accurate, but it could very well be that these
motivations are culture-specific. Therefore, one suggestion for future studies would be to examine whether and how motivational
underpinnings of Big Five personality traits differ from one culture
to another. To address this topic more thoroughly, there is at least
qualitative data on cultural differences in motivational underpinnings needed. For this purpose, cultural anthropologists could be
a great help to psychologists in exploring cross-cultural differences
in the meaning of personality traits.
4.1. Strengths and limitations
Besides the already discussed advantages of the present study,
there are several additional strengths. A first strength concerns
our samples. The advantages of the sizes of our samples has been
described previously, but our samples were also more representative than the college student samples that are typically employed
in cross-cultural research. More specifically, we sampled adolescents representing all various educational levels of high school
available in The Netherlands and Italy. Because high school attendance is obligatory in both these countries, our samples were quite
representative for the general population of adolescents. Previous
studies usually employed college student samples (Schmitt et al.,
2007, 2008) that are not necessarily representative for the general
population. Even though it has been shown that mean-level differences and sex differences found among college students are quite
similar to those found among more representative adults samples
(Costa et al., 2001; McCrae, 2001), future cross-cultural studies
should seek to employ larger and more representative samples,
like the ones employed in the present study.
A second strength concerns the examination of cross-cultural
differences of personality with two distinct types of psychopathological symptoms (i.e., depressive symptoms and the Generalized
Anxiety Disorder symptom of worry). Cross-cultural research on
this topic was lacking, while the present study shows that there
can be differences between countries in such linkages that could
potentially provide insights into the different connotation personality traits may have across the globe.
Besides these strengths, the present study also has several limitations. A first limitation concerns our focus on only two European
countries. In the present study, we tentatively attributed cross-cultural differences in the magnitude of sex differences to masculinity. To be more certain that masculinity indeed plays a role in
295
the magnitude of sex differences, multiple masculine cultures
should be compared to multiple feminine cultures.
Our reliance on a measure with adjectives may be considered a
second limitation. Phrase items may be more appropriate for crosscultural comparisons (McCrae & Costa, 1997), but the content of
such phrase items is often also heavily dependent on adjectives
(Nye et al., 2008). That is, the whole meaning of the phrase is often
predominantly determined by the meaning of the one adjective in
that phrase. Therefore, it is unlikely that we would have found scalar invariance if we would have used phrase items.
Overall, our Italian-Dutch comparison can be considered an
illustration of a more general problem. That is, cross-cultural research may be appealing, but extreme caution needs to be warranted when interpreting the findings of such studies. As
previous cross-cultural research has not always taken the appropriate precautions, we would recommend the use of rigorous tests
of measurement invariance and careful analyses of the meaning of
items in different cultures before proceeding to interpreting crosscultural differences in personality traits.
References
Allik, J., & McCrae, R. R. (2004). Toward a geography of personality traits: patterns of
profiles across 36 cultures. Journal of Cross-Cultural Psychology, 35, 13–28.
American Psychiatric Association (2000). Diagnostic and statistical manual of mental
disorders (4th ed.). Washington, DC: American Psychiatric Association Inc. [Text
Revision (DSM-IV-TR)].
Barbaranelli, C., Caprara, G. V., Rabasca, A., & Pastorelli, C. (2003). A questionnaire
for measuring the big five in late childhood. Personality and Individual
Differences, 34, 645–664.
Birmaher, B., Khetarpal, S., Brent, D., Cully, M., Balach, L., Kaufman, J., et al. (1997).
The screen for child anxiety related emotional disorders (scared): Scale
construction and psychometric characteristics. Journal of the American
Academy of Child and Adolescent Psychiatry, 36, 545–553.
Branje, S. J. T., van Lieshout, C. F. M., & van Aken, M. A. G. (2004). Relations between
big five personality characteristics and perceived support in adolescents’
families. Journal of Personality and Social Psychology, 86, 615–628.
Byrne, B. M., Shavelson, R. J., & Muthen, B. (1989). Testing for the equivalence of
factor covariance and mean structures: The issue of partial measurement
invariance. Psychological Bulletin, 456, 466.
Caprara, G. V., Barbaranelli, C., Bermudez, J., Maslach, C., & Ruch, W. (2000).
Multivariate methods for the comparison of factor structures in cross-cultural
research: An illustration whit the italian Big Five Questionnaire. Journal of Cross
Cultural Psychology, 31, 437–464.
Caprara, G. V., Barbaranelli, C., Pastorelli, C., & Cervone, D. (2004). The contribution
of self-efficacy beliefs to psychosocial outcomes in adolescence. Predicting
beyond global dispositional tendencies. Personality and Individual Differences,
37, 751–763.
Caspi, A., Roberts, B. W., & Shiner, R. L. (2005). Personality development: Stability
and change. Annual Review of Psychology, 56, 453–484.
Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement
invariance. Structural Equation Modeling, 14, 464–504.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for
testing measurement invariance. Structural Equation Modeling, 9, 233–255.
Costa, P. T., Terracciano, A., & McCrae, R. R. (2001). Sex differences in personality
traits across cultures: Robust and surprising findings. Journal of Personality and
Social Psychology, 81, 322–331.
Crocetti, E., Hale, W. W., III, Fermani, A., Raaijmakers, Q. A. W., & Meeus, W. H. J.
(2009). Psychometric properties of the Screen for Child Anxiety Related
Emotional Disorders (SCARED) in the general Italian adolescent population: A
validation and a comparison between Italy and The Netherlands. Journal of
Anxiety Disorders, 23, 824–829.
Crocetti, E., Rubini, M., & Meeus, W. H. J. (2008). Capturing the dynamics of identity
formation in various ethnic groups: Development and validation of a threedimensional model. Journal of Adolescence, 31, 207–222.
Crocetti, E., Schwartz, S. J., Fermani, A., & Meeus, W. H. J. (2010). The Utrechtmanagement of identity commitments scale (U-MICS): Italian validation and
cross-national comparisons. European Journal of Psychological Assessment, 26,
172–186.
Currie, C., Gabhainn, S. N., Godeau, E., Roberts, C., Smith, R., Currie, D., et al. (2008).
Inequalities in young people’s health. Health behaviour in school-aged children
international report from the 2005/2006 survey. Copenhagen: World Health
Organization.
De Fruyt, F., de Bolle, M., McCrae, R. R., Terraciano, A., & Costa, P. T.Collaborators of
the Adolescent Personality Profiles of Cultures Project. (2009). Assessing the
universal structure of personality in early adolescence. The NEO-PI-R and NEOPI-3 in 24 cultures. Assessment, 16, 301–311.
De Raad, B., Barelds, D. P. H., Levert, E., Ostendorf, F., Mlacic, B., Di Blas, L., et al.
(2010). Only three factors of personality description are fully replicable across
Author's personal copy
296
T.A. Klimstra et al. / Journal of Research in Personality 45 (2011) 285–296
languages: A comparison of fourteen trait taxonomies. Journal of Personality and
Social Psychology, 98, 160–173.
Denissen, J. J. A., & Penke, L. (2008). Motivational individual reaction norms
underlying the five-factor model of personality: First steps towards a theorybased conceptual framework. Journal of Research in Personality, 42, 1285–1302.
Ellis, B. B., Becker, P., & Kimmel, H. D. (1993). An item response theory evaluation of
an English version of the Trier Personality Inventory (TPI). Journal of CrossCultural Psychology, 24, 133–148.
Gerris, J. R. M., Houtmans, M. J. M., Kwaaitaal-Roosen, E. M. G., Schipper, J. C.,
Vermulst, A. A., & Janssens, J. M. A. M. (1998). Parents, adolescents and young
adults in dutch families: A longitudinal study. Nijmegen: Institute of Family
Studies University of Nijmegen.
Goldberg, L. R. (1992). The development of markers for the big-five factor structure.
Psychological Assessment, 4, 26–42.
Hale, W. W., III, Klimstra, T. A., & Meeus, W. H. J. (2010). Is the generalized anxiety
disorder symptom of worry just another form of neuroticism? A five-year
longitudinal study of adolescents from the general population. Journal of Clinical
Psychiatry, 71, 942–948.
Hale, W. W., III, Raaijmakers, Q. A. W., Muris, P., & Meeus, W. H. J. (2005).
Psychometric properties of the Screen for Child Anxiety Related Emotional
Disorders in the general adolescent population. Journal of the American Academy
of Child and Adolescent Psychiatry, 44, 283–290.
Heine, S. J., Lehman, D. R., Peng, K., & Greenholtz, J. (2002). What’s wrong with crosscultural comparisons of subjective Likert scales: The reference-group problem.
Journal of Personality and Social Psychology, 82, 903–918.
Hofstede, G. (2001). Culture’s consequences: Comparing values, behaviors, institutions
and organizations across nations (second ed.). Beverly Hills, CA: Sage.
Hogan, R. (1996). A socioanalytic perspective on the five-factor model. In J. S.
Wiggins (Ed.), The five factor model of personality: Theoretical perspectives
(pp. 163–179). New York: Guilford Press.
Huang, C. D., Church, A. T., & Katigbak, M. S. (1997). Identifying cultural differences
in items and traits: Differential item functioning in the NEO Personality
Inventory. Journal of Cross-Cultural Psychology, 28, 192–218.
Johnson, W., Spinath, F., Krueger, R. F., Angleitner, A., & Riemann, R. (2008).
Personality in Germany and Minnesota: An IRT-based comparison of MPQ selfreports. Journal of Personality, 76, 665–706.
Klimstra, T. A., Akse, J., Hale, W. W., III, Raaijmakers, Q. A. W., & Meeus, W. H. J.
(2010). Longitudinal associations between Big Five personality traits and
problem behavior in adolescence. Journal of Research in Personality, 44, 273–284.
Klimstra, T. A., Hale, W. W., III, Raaijmakers, Q. A. W., Branje, S. J. T., & Meeus, W. H. J.
(2009). Maturation of personality in adolescence. Journal of Personality and
Social Psychology, 96, 898–912.
Kline, R. B. (2005). Principles and Practice of Structural Equation Modeling. New York:
Guilford.
Kotov, R., Gamez, W., Schmidt, F., & Watson, D. (2010). Linking ‘‘Big’’ personality
traits to anxiety, depressive, and substance use disorders: A meta-analysis.
Psychological Bulletin, 136, 768–821.
Kovacs, M. (1985). The children’s depression, inventory (cdi). Psychopharmacology
Bulletin, 21, 995–998.
Kovacs, M. (1988) Children’s depression inventory. Questionario di autovalutazione
[Children’s depression inventory. Self-report questionnaire] (Adattamento
italiano a cura di M. Camuffo, R. Cerutti, L. Lucarelli, R. Mayer) [Italian
adaptation by M. Camuffo, R. Cerutti, L. Lucarelli, R. Mayer]. Firenze: O.S.
Organizzazioni Speciali.
Krueger, R. F. (1999). Personality traits in late adolescence predict mental disorders
in early adulthood: A prospective-epidemiological study. Journal of Personality,
67, 39–65.
Lewis-Fernandez, R., & Kleinman, A. (1994). Culture, personality and
psychopathology. Journal of Abnormal Psychology, 103, 67–71.
Little, T. D., Cunningham, W. A., Shahar, G., & Widaman, K. F. (2002). To parcel or not
to parcel: Exploring the question, weighing the merits. Structural Equation
Modeling, 9, 151–173.
Luyckx, K., Soenens, B., & Goossens, L. (2006). The personality-identity interplay in
emerging adult women: Convergent findings from complementary analyses.
European Journal of Personality, 20, 195–215.
MacDonald, K. B. (1995). Evolution, the Five-Factor Model, and levels of personality.
Journal of Personality, 63, 525–567.
MacDonald, K. B. (1998). Evolution, culture, and the five-factor model. Journal of
Cross-Cultural Psychology, 29, 119–149.
Marsh, H. W., & Hau, K.-T. (1999). Confirmatory factor analysis: Strategies for small
sample sizes. In R. H. Hoyle (Ed.), Statisitical strategies for small sample research
(pp. 252–284). Thousand Oaks, CA: Sage.
McCrae, R. R. (2001). Trait psychology and culture: Exploring intercultural
comparisons. Journal of Personality, 69, 819–846.
McCrae, R. R., & Costa, P. T. Jr., (1997). Personality trait structure as a human
universal. American Psychologist, 52, 509–516.
McCrae, R. R., & John, O. (1992). An introduction to the five-factor model and its
applications. Journal of Personality, 60, 175–215.
McCrae, R. R., & Terracciano, A.78 Members of the Personality Profiles of Cultures
Project.. (2005). Universal features of personality traits from the observer’s
perspective: Data from 50 cultures. Journal of Personality and Social Psychology,
88, 547–561.
McCrae, R. R., Terraciano, A., De Fruyt, F., De Bolle, M., Gelfand, M. J., Costa, P. T., et al.
(2010). The validity and structure of culture-level personality scores: Data from
ratings of young adolescents. Journal of Personality, 78, 815–838.
Meeus, W. H. J., van de Schoot, R., Klimstra, T. A., & Branje, S. J. T. (in press).
Personality types in adolescence: Change and stability and links with
adjustment and relationships. A five-wave longitudinal study. Developmental
Psychology.
Muthén, L. K., & Muthén, B. O. (2007). Mplus user’s guide (fourth ed.). Los Angeles,
CA: Muthén & Muthén.
Nye, C. D., Roberts, B. W., Saucier, G., & Zhou, X. (2008). Testing the measurement
equivalence of personality adjective items across cultures. Journal of Research in
Personality, 42, 1524–1536.
Peabody, D., & de Raad, B. (2002). The substantive nature of psycholexical
personality factors: A comparison across languages. Journal of Personality and
Social Psychology, 83, 983–997.
Perugini, M., & Richetin, J. (2007). In the land of the blind, the one-eyed man is king.
European Journal of Personality, 21, 977–981.
Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and standard errors
in covariance structure analysis. In A. Von Eye & C. C. Clogg (Eds.), Latent
variables analysis: Applications for developmental research (pp. 399–419).
Thousand Oaks, CA: Sage Publications.
Schmitt, D. P., Allik, J., McCrae, R. R., Benet-Martínez, V., Alcalay, L., Ault, L., et al.
(2007). The geographic distribution of Big Five personality traits: Patterns and
profiles of human self-description across 56 nations. Journal of Cross-Cultural
Psychology, 38, 173–212.
Schmitt, D. P., Raelo, A., Voracek, M., & Allik, J. (2008). Why can’t a man be more like
a woman? Sex differences in Big Five personality traits across 55 cultures.
Journal of Personality and Social Psychology, 94, 168–182.
Selfhout, M., Burk, W., Branje, S. J. T., Denissen, J. J. A., van Aken, M. A. G., & Meeus,
W. H. J. (2010). Emerging late adolescent friendship networks and big five
personality traits: A social network approach. Journal of Personality, 78,
509–538.
Stark, S., Chernyshenko, O. S., & Drasgow, F. (2006). Detecting differential item
functioning with CFA and IRT: Toward a unified strategy. Journal of Applied
Psychology, 91, 1292–1306.
Tackett, J. L. (2006). Evaluating models of the personality–psychopathology
relationship in children and adolescents. Clinical Psychology Review, 26,
584–599.
Timbremont, B., & Braet, C. (2002). Children’s depression inventory: Nederlandstalige
versie [children’s depression inventory: Dutch version]. Lisse: Swets & Zeitlinger.
Vandenberg, R. J., & Lance, C. E. (2000). A review and synthesis of the measurement
invariance literature: Suggestions, practices, and recommendations for
organizational research. Organizational Research Methods, 3, 4–70.
Watson, D., & Clark, L. A. (1984). Negative affectivity: The disposition to experience
aversive emotional states. Psychological Bulletin, 96, 465–490.
Wood, D., Harms, P., & Vazire, S. (2010). Perceiver effects as projective tests: What
your perceptions of others say about you. Journal of Personality and Social
Psychology, 99, 174–190.